Effects of the Sleep SAAF Responsive Parenting Intervention on Rapid Infant Weight Gain: A Randomized Clinical Trial of African American Families

Abstract

Responsive parenting (RP) interventions reduce rapid infant weight gain but their effect for underserved populations is largely unknown. The Sleep SAAF (Strong African American Families) study is a two-arm randomized clinical trial for primiparous African American mother-infant dyads that compares an RP intervention to a child safety control over the first 16 weeks postpartum. Here we report on intervention effects on rapid infant weight gain and study implementation. Families were recruited from a mother/baby nursery shortly after delivery. Community Research Associates (CRAs) conducted intervention home visits at 3 and 8 weeks postpartum, and data collection home visits at 1, 8, and 16 weeks postpartum. To examine rapid infant weight gain, conditional weight gain (CWG) from 3 to 16 weeks, the primary outcome, and upward crossing of 2 major weight-for-age percentile lines were calculated. Among the 212 mother-infant dyads randomized, 194 completed the trial (92% retention). Randomized mothers averaged 22.7 years, 10% were married, and 49% participated in the Supplemental Nutrition Assistance Program (SNAP). Adjusting for covariates, mean CWG was lower among RP infants (0.04, 95% CI [−0.33, 0.40]) than among control infants (0.28, 95% CI [−0.08, 0.64]), reflecting non-significantly slower weight gain (p=0.15, effect size d=0.24). RP infants were nearly half as likely to experience upward crossing of 2 major weight-for-age percentile lines (14.1%) compared to control infants (24.2%), p=0.09, odds ratio=0.52 (95% CI [0.24, 1.12]). Implementation data revealed that participating families were engaged in the intervention visits and intervention facilitators demonstrated high fidelity to intervention materials. Findings show that RP interventions can be successfully implemented among African American families while suggesting the need for modifications to yield stronger effects on infant weight outcomes.

1. Introduction

Infancy is a critical period for obesity risk, with rapid infant weight gain an important early risk factor for later overweight and obesity (Baird et al., 2005; Ong & Loos, 2006; Woo Baidal et al., 2016). Rapid infant weight gain is more common among African American infants than White infants, contributing to disparities in obesity during early childhood (Isong et al., 2018; Taveras et al., 2010). Given these patterns, prevention programs that can reduce the risk of rapid infant weight gain are attractive, with particular promise for responsive parenting (RP) behaviors in sleep, responsive feeding, and routines (Lumeng et al., 2015). Highlighting this interest, a report by the Robert Wood Johnson Healthy Eating Research expert committee stated that following RP is critical to foster optimal child development and prevent excessive weight gain (Pérez-Escamilla et al., 2017).

The Intervention Nurses Start Infants Growing on Healthy Trajectories (INSIGHT) trial (Paul et al., 2014) aimed to reduce rapid infant weight gain early in life by providing caregiving guidance to first-time, primarily White, middle-class mothers 20 years of age and older. This comprehensive RP intervention was delivered by trained nurses during four home visits at 2, 16, 28, and 40 weeks postpartum. The RP intervention provided guidance for parents on how to respond to their infant’s needs across 4 behavioral states (drowsy, sleeping, fussy, alert) and emphasized an RP framework of parenting that is developmentally appropriate, prompt, and contingent on the infant’s needs (Eshel et al., 2006). At 2 weeks, the RP intervention focused on alternatives to using food to soothe fussy but not hungry infants, age-appropriate sleep guidance such as putting baby to bed drowsy but awake, and the division of feeding responsibility that taught parents to identify hunger and fullness cues to allow infants to determine how much and whether to eat. Relative to a safety control group, infants in the RP group gained weight more slowly from birth to 6 months, had lower weight-for-length percentiles at age 1 year, were less likely to be overweight at age 1 year, and had lower BMI z scores at age 3 years (Paul et al., 2018; Savage et al., 2016).

Despite these promising results and guidance from experts identifying a need to conduct RP studies among low socioeconomic status and racial and ethnic minority groups in the United States (US; Pérez-Escamilla et al., 2017), the generalizability of RP interventions among historically underserved families—particularly African American families—is largely unknown. The notable exception is the Mothers & Others trial (Wasser et al., 2020; Wasser et al., 2017), a six-session infant feeding and care home visiting intervention for non-Hispanic Black mothers and a partner (most commonly the infant’s father or grandmother) that began during the second trimester of pregnancy and continued through 12 months postpartum. That study did not find significant differences between the intervention group and the injury prevention control group in infant weight-for-length at 3, 6, 9, 12, or 15 months (Wasser et al., 2020). Additional research on effective RP interventions that target early rapid weight gain among African Americans and other racial and ethnic minority groups is needed and may curb widening disparities in early childhood obesity.

To address these gaps, we conducted the Sleep SAAF (Strong African American Families) trial for primiparous African American mothers and their newborn infants (Lavner et al., 2019). This two-arm randomized clinical trial adapted the INSIGHT 2-week RP curriculum into a standalone RP intervention that was delivered at 3- and 8-weeks postpartum, and compared it to a child safety control intervention over the first 16 weeks postpartum. The study focused solely on the first 16 weeks postpartum in order to precede the transition to solid foods, which was addressed in a separate intervention session in INSIGHT that was administered at 16 weeks postpartum. The current analysis focuses on the primary outcome of the trial: rapid infant weight gain from 3 to 16 weeks. As described in more detail below, we included two indices of rapid weight gain—conditional weight gain (CWG; Griffiths et al., 2009), as in INSIGHT (Savage et al., 2016), as well as upward crossing of 1 or 2 major weight-for-age percentile lines on the WHO growth charts (Centers for Disease Control and Prevention; World Health Organization) given their clinical relevance. In this initial report we also describe 16-week weight outcomes and trial implementation, including recruitment, retention, participant engagement, and intervention fidelity.

2. Methods

2.1. Participants and Procedures

This study was approved by the Augusta University Institutional Review Board and was registered on www.clinicaltrials.gov (NCT03505203). African American mother-infant dyads were recruited into the Sleep SAAF study from the mother/baby nursery at Augusta University Medical Center (AUMC) in Augusta, GA shortly after delivery (mean infant age at enrollment = 1.5 days). Mothers were told that the purpose of the study was to “learn more about effective ways to increase healthy, safe sleep patterns among African American infants and mothers.” Recruitment began in the spring of 2018 and continued through the spring of 2021, with a recruitment pause from March 9, 2020 – August 31, 2020 due to the COVID-19 pandemic. All newborns delivered at AUMC were screened for eligibility. The project’s recruitment coordinator used the electronic medical records system to pre-screen mothers and infants based on the following inclusion criteria: (1) full-term infants (≥ 37 weeks gestational age); (2) singleton infant; (3) infant ≥ 2500 g at birth; (4) primiparous mother ≥ 17 years of age; (5) mother self-identifying as African American; (6) residence ≤ 75 miles from Augusta; and (7) English speaking. In addition, mother-infant dyads were excluded if there was the following: medical condition affecting the infant’s feeding or growth (e.g., cleft palate, complex congenital heart disease); major maternal morbidities or pre-existing conditions affecting postpartum care or the mother’s ability to care for her infant (e.g., drug use, mental health concerns, in-hospital social service referral); plan for the infant to be adopted; or intent to move from the area within four months. The study protocol can be found elsewhere (Lavner et al., 2019).

Consent was obtained by the project’s recruitment coordinator in the hospital. Mother-infant dyads were then visited at home at approximately 1, 3, 8, and 16 weeks postpartum (see Figure 1) by trained Community Research Associates (CRAs) from the Center for Family Research at the University of Georgia. CRAs were African American community members from local communities who received 30+ hours of initial training and ongoing supervision in either data collection or intervention delivery. To minimize bias during the assessments, separate CRAs served as data collectors (at 1, 8, and 16 weeks) and intervention facilitators (at 3 and 8 weeks). Further, intervention facilitators were trained in (and subsequently delivered) only one of the two study interventions (RP or safety control). Participating mothers received $50 after the 1-week visit, $50 after the 3-week visit, $100 after the 8-week visits, and $100 after the 16-week visit.

Figure 1.

Study CONSORT diagram.

Notes. The number screened for eligibility only includes primiparous African American mothers; mothers who were another race or multiparous were not recorded. Mother/newborn pairs meeting more than one exclusion criteria are listed only once in the exclusion breakdown. RP = Responsive Parenting. C19 = COVID-19 pandemic.

Prior to the onset of the COVID-19 pandemic in March 2020, all study visits occurred face-to-face inside participants’ homes. At the onset of the COVID-19 pandemic, participants who were enrolled but who had not yet completed their first intervention visit were withdrawn (n = 6). Follow-up visits for participants who had completed at least one intervention visit were put on hold until summer 2020. At that time, still-enrolled mothers completed self-report measures remotely (n = 23), and a subset of these weighed their infant using study scales (Medela BabyWeigh II Scale, McHenry, Il) that were delivered to their homes with instructions from study staff (n = 12). These visits occurred beyond the typical 16-week timeline (mean infant age = 29 weeks). Beginning September 1, 2020, recruitment resumed. Home visits also resumed for newly enrolled participants, with modified protocols to provide minimal face-to-face contact in line with IRB guidelines. Most notably, intervention visits were delivered via Zoom from outside the participant’s home, with infant weight and length data collected by the CRA in the participant’s home.

2.2. Randomization

Participants were randomized to either the RP or safety control group after completion of the 1-week data collection visit. The randomization scheme stratified on sex-specific birth weight for gestational age (<50th percentile or > 50th percentile) and intended feeding mode (breastfeeding or formula), consistent with INSIGHT (Paul et al., 2014).

2.3. Intervention Groups

The RP intervention provided guidance on RP in the context of sleeping, crying, feeding, and interactive play, drawing from the INSIGHT 2-week RP curriculum (Paul et al., 2014). Guidance on infants’ sleep included establishing a consistent bedtime routine; putting the infant to bed early, drowsy but awake; and avoiding feeding the infant to sleep or putting the infant to bed with a bottle. Guidance on crying included information on reasons for infant crying, that crying does not always indicate hunger, how to discriminate hunger from other reasons for infant crying, and how to use alternative soothing strategies rather than feeding. Mothers were also taught the “5 S’s” soothing strategies (Karp, 2006) that they could use to soothe their crying baby: Shushing, Swinging, Side/Stomach Position, Sucking, and Swaddling. Strategies for dealing with night waking to promote self-soothing were also highlighted, including allowing a brief time for the infant to self-soothe before the mother intervening to soothe the infant. RP facilitators also taught mothers to recognize hunger cues (rooting, mouthing, bringing hand to mouth) and fullness cues (letting go of nipple, falling asleep, turning head away, interest in other things). Mothers were given education on age-appropriate bottle sizes, breast milk/formula volumes, use of slow-flow bottle nipples for infants under 4 months to prevent overfeeding or choking, and how to use fullness cues (rather than the amount of milk in the bottle) to determine when to terminate a feeding. Mothers were also advised that breast milk or formula is best for infants of this age, to delay the introduction of other beverages and solid foods until age 6 months, and to avoid adding infant cereal to a bottle. Lastly, RP facilitators encouraged mothers to play with their baby every day and to have “tummy time” with their baby for at least a few minutes each day. All topics were discussed in detail at the first intervention visit at 3 weeks postpartum (which lasted approximately 90-120 minutes on average), during which mothers were provided with a packet of intervention handouts that the RP facilitators talked through with the mother, supplemented with hands-on activities (e.g., demonstrations of swaddling and side/stomach position), discussion, and videos. The same material was reviewed at the booster intervention visit at 8 weeks postpartum (which lasted approximately 45-60 minutes on average).

Sleep SAAF’s control group received a developmentally-appropriate child safety intervention at 3 and 8 weeks. Visits included information on Sudden Infant Death Syndrome (SIDS) facts and myths, reducing baby’s risk, and the importance of good health care; taking care of a crying baby; finding caretakers for baby; and food safety, including formula and bottle handling, preparation, and storage. Additional information included home safety tips such as preventing falls, water safety, poison prevention, fire safety, and preventing burns, as well as car seat safety. The home visits were designed to be matched in length and intensity to the RP intervention visits and to avoid messages related to RP. As in the RP condition, safety mothers received a packet of safety handouts which the safety facilitators talked through with the mother, supplemented with hands-on activities (e.g., reviewing child safety websites), discussion, and videos.

Several efforts were made to ensure the integrity of the two intervention conditions. Both conditions included detailed and scripted manuals of intervention content, standardized training procedures, evaluation of the facilitators’ delivery of the curriculum materials, and regular monitoring in the field. Audio recordings of scheduled visits were reviewed by the study’s intervention coordinator, and feedback was provided individually and during group trainings to minimize drift in the facilitators’ skills and to enhance their ability to implement the curriculum as intended.

Methods and intervention materials for both groups were tailored for African American families. Study design elements to enhance the program’s ability to recruit and retain African American families included eligibility regardless of feeding mode, the emphasis on healthy sleep during recruitment (rather than an explicit focus on feeding practices and/or infant weight) based on feedback from a focus group conducted prior to the study, the use of home visits to reduce barriers to participation, and implementation by CRAs. Additionally, intervention materials featured photographs of African American mothers and their infants and incorporated safe sleep strategies from a program to prevent SIDS among African American families published by the National Institute of Child Health and Human Development (NICHD, 2006). Additional detail about the RP and safety control interventions is provided in Lavner et al. (2019).

2.4. Measures

2.4.1. Background Characteristics

Maternal age, self-identified race, and infant gestational age were extracted from electronic medical records. Mothers’ height (Seca 274, Hanover, MD) was measured in duplicate by trained research staff at enrollment, who also collected mothers’ self-reported pre-pregnancy weight. Family demographic information was collected from mothers at the first data collection visit at 1 week postpartum.

2.4.2. Infant Growth

Research staff measured infant weight (Medela BabyWeigh II Scale, McHenry, Il) at each visit (enrollment, 1, 3, 8, and 16 weeks) in duplicate (or triplicate if the first two measurements differed by more than 50 g). Infant recumbent length (Seca 416 Infantometer, Hanover, MD) was measured at enrollment and 16 weeks in duplicate (or triplicate if the first two measurements differed by more than 0.2 cm). The weight and length protocols followed standardized procedures (Centers for Disease Control and Prevention). Anthropometry measurements were recorded on iPads by research staff using Qualtrics, a secure, web-based survey interface. Infant measurements were used to derive weight-for-age at enrollment, 3, 8, and 16 weeks and length-for-age, weight-for-length and BMI-for-age at enrollment and 16 weeks, which were converted to z scores and percentiles using World Health Organization (WHO) reference standards (de Onis et al., 2004).

2.4.3. Conditional Weight Gain

Adapting the methods from Griffiths et al. (2009) and consistent with our previous work in INSIGHT (Savage et al., 2016), conditional weight gain (CWG) scores were calculated as the standardized residuals from the linear regression of weight-for-age z scores at 16 weeks on weight-for-age scores at 3 weeks, length-for-age z scores at enrollment and 16 weeks, and infant age at 16 weeks. CWG scores represent the remaining variation in infant weight gain that is not explained by weight at 3 weeks (the first intervention visit), length at enrollment, length at 16 weeks, or actual child age at the 16-week visit. A CWG score of zero represents the sample mean, with positive scores indicating more rapid or faster than average infant weight gain and negative scores indicating slower infant weight gain (Griffiths et al., 2009). In INSIGHT, CWG from birth to 28 weeks was positively associated with BMI z score, overweight, and obesity at 3 years (Paul et al., 2018).

2.4.4. Upward Crossing of Major Percentile Lines

Rapid infant weight gain was also examined as an upward crossing of 1 and 2 major weight-for-age percentile lines from 3 to 16 weeks (Taveras et al., 2011) using the WHO weight-for-age growth charts from the CDC (Centers for Disease Control and Prevention; World Health Organization). The major percentile lines are the 2^nd, 5^th, 10^th, 25^th, 50^th, 75^th, 90^th, 95^th, and 98^th.

2.4.5. Growth-Related Adverse Events

Growth-related adverse events included weight-for-age below the 5^th percentile and/or downward crossing of 2 major weight-for-age percentile lines between study visits on the WHO weight-for-age growth charts from the CDC (Centers for Disease Control and Prevention; World Health Organization).

2.4.6. Treatment Fidelity

All intervention visits at 3 and 8 weeks were audio recorded. A random sample of RP (n = 43) and safety control (n = 41) recordings were coded by a trained rater using detailed checklists for adherence to program guidelines. All facilitators were assessed at least once. Of the audio recordings reviewed, approximately 20% (n = 19) were coded by a second rater to ensure consistency. The intraclass correlation coefficient between raters was 0.99. Facilitators also provided their own ratings of intervention fidelity after each session, responding to the prompt “I was able to cover key messages (as written) with the mother.” Responses were 1 = implemented fully as written; 2 = implemented with minor modifications (adjustments to or condensing of 1 or 2 elements); and 3 = implemented with notable modifications (adjusting/condensing 3 or more elements or skipping more than 1 element).

2.4.7. Maternal Session Engagement

After the intervention visits at 3 and 8 weeks, facilitators rated mothers’ engagement by responding to the following prompt: “Which of the following choices best describes the mother’s participation, motivation, and communication?” Responses were 1 = High engaged: Demonstrated a high level of active engagement for all of session (90-100%); appeared highly motivated to listen; attempted to communicate his/her understanding and/or impressions about the lessons; 2 = Moderately engaged: Showed high levels of engagement and communication for most of the session (75-90%); disengaged or less communicative at some times; parent demonstrated a consistent, but only moderate level of engagement and communication (e.g., participated passively, or with lower levels of verbal communication); 3 = Somewhat engaged: Struggled to maintain positive and active engagement during the session; difficulty grasping the concepts or apparent disinterest; with staff support, participated actively in most of the session (50-75%); and 4 = Rarely engaged: Parent did not sustain active engagement during this session (0-50%); tended to remain disinterested or disengaged for some substantial portion of the session.

2.5. Sample Size and Power Calculations

The primary outcome for Sleep SAAF is between-group differences in infant CWG scores from 3 to 16 weeks. Our target enrollment was 300 randomized mother-infant dyads based on the INSIGHT effect size of approximately 0.4 for infant CWG at 6 months (Savage et al., 2016). Specifically, we used G Power 3.1.9.2 (Faul et al., 2007) to estimate the smallest intervention effect size detectable with 80% power, a 5% two-tailed Type 1 error rate, and two groups of 150 each (assuming 10% attrition over the course of the study). For this target sample of 270 completed, we would have achieved 80% power with an effect size d of 0.34.

Enrollment ended in spring 2021 before this target was met due to the aforementioned challenges from the COVID-19 pandemic and the funding timeline. The final sample included 212 randomized mother-infant dyads, of whom CWG scores were available for 171 infants. This sample size resulted in 80% power to detect an effect size d of 0.43 with a 5% two-tailed Type 1 error rate.

2.6. Analysis Plan

Descriptive statistics including means and standard deviations (SD) and frequencies (percentages) were used to summarize key study variables. General linear models examined the effect of the RP intervention (relative to the safety control intervention) on infant CWG from 3 to 16 weeks. Adjusted models included maternal age, pre-pregnancy BMI, gestational weight gain, highest education level, participation in the Supplemental Nutrition Assistance Program (SNAP), participation in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), infant gestational age, breastfeeding at 3 weeks (any breastfeeding vs. exclusive formula), and study stage (completed pre-COVID or during COVID) as covariates. Effect of study group on upward crossing of 1 and 2 major weight-for-age percentile lines (yes/no) was tested using logistic regression. Independent t-tests examined group differences in infant weight measures (weight-for-age z scores, weight-for-length z scores, and BMI-for-age z scores) at 16 weeks. Lastly, to identify potential differences in intervention efficacy based on infant and/or maternal factors, exploratory subgroup analyses for 16-week weight-for-length z scores examined intervention efficacy by: 1) male vs female children; 2) maternal pre-pregnancy overweight vs not; 3) received SNAP benefits vs not; 4) exclusive breastfeeding at 3 weeks vs not; and 5) maternal age older than 18 vs 18 and younger.

SPSS Version 27 (IBM Corp, Armonk, NY) and SAS 9.4 (SAS Institute, Cary NC) were used to analyze all data and statistical significance was defined a priori as p < 0.05 (two-tailed). Effect sizes were calculated as the mean difference from RP to control groups of continuous weight outcome measures, divided by the root mean square error (pooled variance) from the analyses. For discrete outcomes, logistic regression parameter estimates and Wald chi-squares are reported as the effect and test-statistic respectively.

3. Results

Mothers were African American/Black (100%), non-Hispanic (99%), married (10%) or living with a romantic partner (31%), working full- or part-time prior to birth (48%), some high school or completed high school (62%), and received SNAP (49%) and WIC (76%). Over half of mothers (52.7%) reported any breastfeeding at the time of the first intervention visit (3 weeks postpartum). Demographic characteristics by study group are presented in Table 1.

Table 1.

Demographic characteristics of 212 mother-infant dyads who were randomized to the responsive parenting (RP) and safety control groups.

	Total (n=212)	RP (n=108)	Safety Control (n=104)	Chi sq or test	Prob > value
Infant
Male, n (%)	102 (48%)	53 (49%)	49 (47%)	0.08	0.78
Gestational age (weeks), mean (SD)	39.1 (1.0)	39.1 (1.1)	39.1 (1.0)	−0.25	0.81
Enrollment weight (kg), mean (SD)	3.02 (0.37)	3.06 (0.38)	2.99 (0.36)	−1.33	0.18
Enrollment length (cm), mean (SD)	48.7 (1.7)	48.8 (1.8)	48.7 (1.6)	−0.32	0.75
3-week weight (kg), mean (SD)	3.69 (0.45)	3.71 (0.46)	3.67 (0.45)	−0.54	0.59
3-week any breastfeeding (%)	107 (52.7%)	55 (53.9%)	52 (51.5%)	0.12	0.73
Mother
Age (years), mean (SD)	22.7 (4.5)	23.4 (4.9)	22.0 (3.9)	−2.33	0.02
Prepregnancy BMI, mean (SD)	28.1 (8.2)	27.8 (8.5)	28.4 (7.9)	0.55	0.58
Non-Hispanic, n (%)	208 (99%)	106 (98%)	102 (99%)	0.29	0.59
Maternal pre-pregnancy weight status				2.17	0.54
Underweight (Pre-preg BMI < 18.5)	9 (5%)	6 (6%)	3 (3%)
Normal (Pre-preg BMI 18.5-25)	79 (40%)	43 (42%)	36 (38%)
Overweight (Pre-preg BMI 25-30)	35 (18%)	15 (15%)	20 (21%)
Obese (Pre-preg BMI ≥ 30)	75 (38%)	38 (37%)	37 (39%)
Gestational weight gain (kg), mean (SD)	14.8 (9.0)	15.9 (9.4)	13.7 (8.4)	−1.73	0.08
Diabetes during pregnancy, n (%)	14 (7%)	6 (6%)	8 (8%)	0.56	0.97
Smoked during pregnancy, n (%)	7 (3%)	1 (1%)	6 (6%)	3.89	0.05
Romantic status, n (%)				0.12	0.39
Single	86 (41%)	42 (39%)	44 (42%)
Married/living together	22 (10%)	15 (14%)	7 (7%)
Married, but not living with partner	0 (0%)	0 (0%)	0 (0%)
Living together	66 (31%)	33 (31%)	33 (32%)
Involved/steady/Not living together	37 (18%)	17 (16%)	20 (19%)
Involved/on-off again relationship	1 (1%)	1 (1%)	0 (0%)
Annual household income, n (%)				15.49	0.03
<$10,000	48 (23%)	28 (26%)	20 (19%)
$10,000-$24,999	25 (12%)	8 (8%)	17 (16%)
$25,000-$49,999	29 (14%)	19 (18%)	10 (10%)
$50,000 or more	16 (8%)	8 (8%)	8 (8%)
Don’t know	84 (40%)	38 (36%)	46 (44%)
Refuse to answer	9 (4%)	6 (6%)	3 (3%)
Education, n (%)				2.04	0.73
Some high school (9-11)	30 (14%)	15 (14%)	15 (14%)
High school graduate	102 (48%)	50 (46%)	52 (50%)
Some college or technical school	53 (25%)	27 (25%)	26 (25%)
Completed college	19 (9%)	10 (9%)	9 (9%)
Post graduate training degree	8 (4%)	6 (6%)	2 (2%)
Federal Nutrition Assistance, n (%)
SNAP participation (yes)	101 (49%)	53 (50%)	48 (48%)	0.18	0.67
WIC participation (yes)	159 (76%)	84 (80%)	75 (73%)	1.49	0.22

Abbreviation: BMI = body mass index (calculated as weight in kilograms divided by height in meters squared). RP = responsive parenting. Mean infant age at enrollment was 1.5 days.

3.1. Participant Recruitment and Retention

The study CONSORT diagram is provided in Figure 1. Most of the sample (77%, n = 149) completed the study prior to the onset of the COVID-19 pandemic, with 12% (n = 23) beginning it prior to the pandemic and completing it during the pandemic, and the remaining 11% (n = 22) enrolled during the pandemic when the study resumed with modified protocols in September 2020. Most (80%) eligible families who were approached for the study enrolled. Of those who enrolled, 91% participated in the first data collection visit at 1 week postpartum and were randomized to either the RP group or the safety control group. Of the randomized sample, 92% completed the 16-week data collection visit. Completion rates at 16 weeks did not significantly differ between participants randomized to the RP group (89%) and those randomized to the safety control group (94%), χ² = 1.95, p = 0.16.

3.2. Primary Outcome of Rapid Infant Weight Gain

3.2.1. Conditional Weight Gain from Age 3 to 16 Weeks by Study Group

As shown in Table 2, mean CWG was negative for infants in the RP group (M = −0.09, 95% CI [−0.29, 0.13]), reflecting a slower-than-average pattern of weight gain. In contrast, mean CWG was positive for infants in the safety control group (M = 0.09, 95% CI [−0.13, 0.30]), reflecting faster-than-average weight gain. Group differences without covariates did not reach statistical significance (p = 0.27), effect size d = .18. A slightly stronger pattern of results emerged, although still non-significant, after adjusting for covariates: RP CWG: M = 0.04, 95% CI [−0.33, 0.40], control CWG: M = 0.28, 95% CI [−0.08, 0.64], p = 0.15, effect size d = 0.24. None of the covariates significantly moderated the effect of study group on CWG (data not shown).

Table 2.

The effect of study group on mean conditional weight gain (CWG) from 3 to 16 weeks¹ before (Model 1) and after adjusting for covariates² (Model 2).

	RP Mean [95% CI]	Safety Control Mean [95% CI]	F-value	Prob>F	Effect size3
Model 1: Unadjusted (n = 171)
CWG	−0.09 [−0.29, 0.13]	0.09 [−0.13, 0.30]	1.21	0.27	0.18
Model 2: Adjusted for Covariates (n=161)
CWG	0.04 [−0.33, 0.40]	0.28 [−0.08, 0.64]	2.09	0.15	0.24

¹CWG scores calculated as studentized residuals from the model Weight-for-age_z_16w = Weight-for-age_z_3w + Length-for-age_z_enrollment + Length-for-age_z_16w + infant age_16w

²Covariates included: maternal pre-pregnancy body mass index (BMI) (kg/m²), maternal age (years), gestational weight gain (kg), maternal education level, participation in WIC (y/n), receipt of SNAP benefits (y/n), gestational age (weeks), breastfeeding at 3 weeks (any breastfeeding vs exclusive formula), and study stage (completed pre-COVID or during COVID). All covariates were not statistically significant p > 0.05.

³Effect size was calculated as the mean difference between Responsive Parenting (RP) and Safety Control groups divided by the pooled mean square error.

3.2.2. Upward Crossing of Percentile Lines from Ages 3 to 16 Weeks by Study Group

Examination of upward crossing of 2 major weight-for-age percentile lines from 3 to 16 weeks showed that 14.1% of RP infants (n = 13) compared to 24.2% of control infants (n = 22) experienced upward crossing of 2 percentile lines (p = 0.09). The odds ratio for this comparison was 0.52 (95% CI [0.24, 1.12]), meaning RP group infants were 0.52 times less likely than control group infants to upward cross 2 percentile lines from 3 to 16 weeks. Examination of upward crossing of 1 major weight-for-age percentile line from 3 to 16 weeks showed that 48.9% of RP infants (n = 45) compared to 61.5% of control infants (n = 56) experienced upward crossing of 1 percentile line over this period (p = 0.09), odds ratio = 0.60 (95% CI [0.33, 1.11]).

3.3. Secondary Outcomes: Weight at Age 16 Weeks by Study Group

At 16 weeks, there were no significant differences between infants in the RP and control groups for weight-for-age percentiles or z scores, weight-for-length percentiles or z scores, or BMI-for-age percentiles or z scores, with effect size ds ranging from 0.00 to 0.10 (see Table 3). Exploratory subgroup analyses on 16-week weight-for-length z scores examining whether the RP intervention was more effective for certain participants revealed only one significant effect. Specifically, for mothers who did not receive SNAP benefits, there was a significant effect of condition (p = 0.04): the mean for the safety control group (n = 44) was 0.05 (95% CI [−0.26, 0.35]) and the mean for the RP group (n = 41) was −0.43 (95% CI [−0.78, −0.08]), indicating significantly lower weight-for-length among RP infants whose mothers did not receive SNAP benefits. For infants whose mothers received SNAP benefits, there was no significant effect for condition (p = 0.14). No other subgroups exhibited significant condition effects in 16-week weight-for-length z scores (data not shown).

Table 3.

Weight outcomes at 16 weeks by study group.

	RP	Safety Control
Upward crossing of percentile lines on the WHO growth chart, based on weight-for-age from 3 to 16 weeks
	N (%)	N (%)	Wald χ2	Pr>ChiSq	Effect
Crossing 2 percentile lines (n = 183)	13 (14.1%)	22 (24.2%)	2.9	0.09	0.33
Crossing 1 percentile line (n = 183)	45 (48.9%)	56 (61.5%)	2.9	0.09	0.26
Growth measurement at 16 weeks, adjusted for WHO standards
	Mean (SD)	Mean (SD)	t-statistic	p-value	Effect size
Weight-for-age percentile (n=183)	41.0 (26.5)	43.9 (30.0)	0.68	0.50	0.10
Weight-for-age z score (n=183)	−0.27 (0.90)	−0.22 (1.03)	0.38	0.71	0.05
Length-for-age percentile (n=171)	49.6 (27.7)	50.5 (28.5)	0.21	0.83	0.03
Length-for-age z score (n=171)	−0.01 (0.93)	−0.00 (0.94)	0.07	0.95	0.00
Weight-for-length percentile (n=171)	41.0 (27.4)	41.4 (29.3)	0.10	0.92	0.01
Weight-for-length z score (n=171)	−0.33 (1.03)	−0.31 (1.13)	0.13	0.89	0.02
BMI percentile (n=171)	39.0 (26.6)	39.9 (29.2)	0.22	0.83	0.03
BMI z score (n=171)	−0.38 (0.98)	−0.35 (1.11)	0.15	0.88	0.03

3.4. Implementation Outcomes

3.4.1. Treatment Fidelity

Facilitator fidelity to RP and safety control guidelines was first examined using the sample of coded audio recordings. For the RP group, mean fidelity adherence across facilitators was 90% (SD = 7.3%) at 3 weeks and 88% (SD =7.9%) at 8 weeks. For the safety control group, mean fidelity adherence across facilitators was 97% (SD = 3.7%) at 3 weeks and 95% (SD = 4.9%) at 8 weeks. Next, we examined facilitators’ self-reported fidelity ratings. For the RP group, nearly all facilitators reported that they implemented the intervention “fully as written” or with only “minor modifications” at 3 (97%) and 8 weeks (93%). High treatment fidelity was also reported by facilitators in the safety control group at 3 (98%) and 8 weeks (99%).

3.4.2. Maternal Session Engagement

Facilitators in both groups reported favorable levels of maternal engagement. In the RP group, 93% of mothers at 3 weeks and 92% of mothers at 8 weeks were rated as having “high” or “moderate” levels of engagement. In the safety control group, 97% of mothers at 3 weeks and 98% of mothers at 8 weeks were rated as having high or moderate levels of engagement.

3.4.3. Growth-Related Adverse Events

Thirty-two infants (15%) met at least one of the two growth monitoring criteria over the course of the study. For weight-for-age below the 5^th percentile, 12% of infants in the RP group (n = 13) and 14% of infants in the safety control group (n = 15) met criteria, a non-significant difference between groups, χ² = 0.26, p = 0.61. For downward crossing of 2 major percentile lines, 6% of RP infants (n = 7) and 5% of control infants (n = 5) met criteria, a non-significant difference between groups, χ² = 0.28, p = 0.60.

4. Discussion

The Sleep SAAF study adapted the INSIGHT 2-week RP curriculum that showed promise in a primarily White, middle-class sample (Savage et al., 2016), for an exclusively African American sample living in the southeastern US. Centuries of systemic racism and oppression have resulted in African American and Black mothers in this region having among the highest poverty rates in the US (Burton et al., 2017). Sleep SAAF families were similarly characterized by lower levels of household income relative to INSIGHT. Sleep SAAF mothers were also younger, had lower levels of education, and were less likely to report being married or in a romantic relationship than the mothers in INSIGHT. Results revealed that the RP intervention was implemented successfully by CRAs and had small, non-statistically significant effects on rapid infant weight gain in the predicted direction as measured by CWG scores and upward major percentile crossing for weight-for-age. Further, there was no evidence that the RP intervention led to insufficient weight gain relative to the safety control group, reducing concerns about potential adverse effects of the RP intervention among this population.

Although our responsive parenting intervention did not significantly impact the primary outcome of infant CWG, findings were in the expected direction and consistent with our hypothesis and previous studies. Infants in the RP group demonstrated slower weight gain relative to infants in the control group, with an effect size d of .24, and were nearly half as likely to experience upward crossing of major weight-for-age percentiles than control infants. These results are consistent with those from our earlier trials indicating that RP infants experienced less rapid weight gain than safety control infants (Paul et al., 2011; Savage et al., 2016). The only other study to focus on early obesity prevention among African American infants did not find significant intervention effects on weight outcomes (Wasser et al., 2020), which is in line with other research examining infant weight effects for programs targeting underserved and under-resourced populations in the US (Kavanagh et al., 2008; Reifsnider et al., 2018). The uptake and efficacy of RP interventions for African American families could be affected by the difficult social context that many African Americans must navigate due to historical legacies and modern manifestations of systemic racism and oppression, and by African American parents reporting greater use of feeding practices that affect obesity risk [e.g., less breastfeeding (Li et al., 2019), earlier introduction of solid foods (Taveras et al., 2010), putting cereal in the infant’s bottle (Berger et al., 2017), more pressuring the infant to eat and using feeding to soothe (Berger et al., 2017)].

There are several other possible explanations for Sleep SAAF having smaller, non-significant effects on rapid infant weight gain relative to INSIGHT. First, although we had high retention rates (92%) of randomized participants, the number of enrolled participants (212 randomized, 194 completed the study) was lower than planned (300 randomized, 270 study completers based on anticipated 10% dropout rate). We did not meet this target due to challenges resulting from the COVID-19 pandemic, which disrupted many clinical trials (McDermott & Newman, 2021). In Sleep SAAF, recruitment was discontinued for nearly 6 months and length data could not be collected for the subset of families who were enrolled when the pandemic began (and thus they could not be included in the CWG analyses). Further, enrollment rates were lower once the study resumed with modified protocols in September 2020 (87% of eligible families enrolled pre-pandemic vs. 49% of eligible families enrolled post-pandemic), potentially due to the pandemic’s disproportionate impact on the health and well-being of Black Americans (Adesogan et al., 2021). A larger sample size would have provided greater power to detect the small effects found here and allowed for more robust analysis of moderation and subgroup effects.

Second, study design features may partially explain differences in RP intervention efficacy on rapid infant weight gain compared to that observed in the INSIGHT trial. It is possible that the 16-week assessment was not long enough for group differences in weight to emerge. We set this window to precede the transition to solid foods, but in doing so may have pre-empted the potential to observe stronger effects later in infancy. In INSIGHT, we found evidence that the RP intervention was associated with lower CWG scores from birth to 28 weeks (Savage et al., 2016), and other studies examining prevention of rapid infant weight gain have focused on changes from birth to 6 or 12 months (Rotevatn et al., 2020). A longer follow-up period would have allowed for more time for the behavioral changes targeted by the intervention to influence, and consequently result in group differences in, infant weight. In addition, the eligibility criteria differed between studies. For example, INSIGHT mothers had to be older (≥20 years old) than Sleep SAAF mothers (≥17 years old), which may have influenced intervention uptake and use of RP messaging. Future RP trials in African American families with longer-term follow-up are needed given that the observed findings of the RP intervention on rapid infant weight gain were in the expected direction and had a small effect size.

Lastly, the RP intervention was brief, with an intervention visit at 3 weeks postpartum that included all intervention material and a booster visit at 8 weeks postpartum to reinforce key RP messages. It is possible that more sessions were needed to facilitate intervention uptake, either by starting prenatally or by adding more intervention visits beyond 8 weeks. In INSIGHT, the focus of the intervention was expanded after 16 weeks to include repeated exposure to promote acceptance of new foods, serving age-appropriate foods and portion sizes, and establishing feeding and sleep routines with clear expectations (Paul et al., 2014). In addition, INSIGHT used color-coded growth charts to provide parents with guidance on typical patterns of child growth and weight gain and feedback based on the individual child’s anthropometrics. Similar additions may be needed in Sleep SAAF to fully realize the potential benefits of the early RP intervention. Future RP interventions may also consider incorporating other caregivers, as in the Mothers & Others trial (Wasser et al., 2017), or offering peer mentoring to increase coordination of and social support around intervention messaging.

There were several promising indicators regarding study implementation. Eligible participants enrolled at high rates (80%), exceeding that in INSIGHT (45%). Additionally, of the randomized sample, 92% completed data collection at 16 weeks, paralleling that in INSIGHT (92%). Participating mothers were rated by intervention facilitators as being engaged in the intervention visits, with more than 90% of mothers in both conditions rated as having high or moderate levels of engagement. These patterns indicate that African American mothers were highly interested and willing to participate in this type of intervention study, and remained engaged once enrolled. These data are particularly promising in light of the continued need for research that reduces health disparities and promotes health equity for racial and ethnic minorities, individuals living with low socioeconomic status, and other underrepresented and underserved populations (NIH, 2021).

The high enrollment and engagement rates in Sleep SAAF may have been enhanced by several study design decisions that we believe increased the study’s appeal and acceptability. Notably, we chose CRAs as intervention facilitators based on other studies led by our team at the Center for Family Research showing that the use of CRAs was a successful staffing model in intervention trials for African American families with older children (Barton et al., 2018; Brody, 2016). CRAs were African American women from similar communities as participants, but did not have specialized knowledge in infant health and development. This marked a change from INSIGHT, which used nurses, and differed from the only other trial focused specifically on African American mothers (Mothers & Others; Wasser et al., 2017), which used Peer Educators who had an MS/MPH degree in a health-related field or a BS/BA degree in a health-related field and 2+ years of experience. Our data indicated that the CRAs were able to implement the RP intervention with high fidelity, based both on their own ratings after each intervention session as well as objective ratings from trained coders. These findings indicate that the RP intervention was generally delivered as intended, providing greater confidence in study group comparisons. The high fidelity demonstrated by the CRAs also supports the rationale that the RP intervention can be delivered successfully by individuals without a nursing or health-related background, increasing the potential for large-scale implementation. Other factors that may have strengthened mothers’ interest in the Sleep SAAF trial include: (1) framing the study as primarily being about promoting healthy infant sleep rather than focusing on infant weight and feeding practices, (2) enrolling mother-infant dyads regardless of their feeding mode, (3) in-home delivery of study assessments and interventions, and (4) tailoring of intervention materials for African American families.

Study limitations must be acknowledged. First, our sample was limited to primiparous African American mothers recruited from a single hospital in the southeastern US. It will be important to examine how findings generalize to other samples of primiparous African American mothers and to multiparous African American mothers. Second, although facilitators rated mothers’ engagement in the intervention sessions, we do not have reports on mothers’ self-reported engagement or their adherence to intervention material after these sessions. Given that between-group differences in infant weight would likely only arise from changes in parent and infant behaviors targeted by the RP intervention, it will be important to examine how the intervention affected sleep, soothing, and feeding. This ongoing work will provide more insights into the degree to which participating mothers in the RP group implemented intervention guidelines and highlight the full set of potential benefits from the RP intervention. Third, a portion of the sample participated in the study with a modified protocol due to the COVID-19 pandemic that included delivery of the intervention via Zoom rather than in-person. Our analyses indicated that study phase did not significantly predict outcomes (or significantly moderate intervention effects), raising the possibility that it may be feasible to deliver the RP intervention in this format in the future. Nonetheless, within-sample variability increased. Lastly, participants in the control condition also received an intervention. The child safety-focused intervention was not intended to affect infant weight or RP, but it is possible that the added attention of home visits influenced outcomes relative to what would have been observed in a no-treatment control condition.

5. Conclusion

These initial findings from the Sleep SAAF trial suggest that a brief, culturally-adapted RP intervention for African American mothers and their infants can be implemented early in life with high engagement from families and high fidelity from trained community members. Our success in doing so is particularly notable given that African American families and other under-resourced and marginalized populations have been underrepresented in previous trials of RP interventions targeting early obesity prevention. There was evidence of small, albeit not statistically significant, differences between RP and control infants in rapid weight gain, extending the findings from our earlier trials. Ongoing work from the Sleep SAAF trial examining group differences in sleep, soothing, and feeding outcomes will advance understanding of the RP intervention’s effects and inform future studies aimed at promoting healthy development and preventing obesity among African American children.

List of abbreviations:

AUMC

Augusta University Medical Center

CRA

Community Research Associate

CWG

Conditional weight gain

INSIGHT

Intervention Nurses Start Infants Growing on Healthy Trajectories

NICHD

National Institute of Child Health and Human Development

RP

Responsive parenting

RWG

Rapid weight gain

SAAF

Strong African American Families

SIDS

Sudden Infant Death Syndrome

SNAP

Supplemental Nutrition Assistance Program

WIC

Special Supplemental Nutrition Program for Women, Infants, and Children